It may be true, as F. Scott Fitzgerald once said, that there are no second acts in American lives. But that definitely isn't the case with American medicines.

Take, for example, the antibiotic neomycin.

Fifty years ago, neomycin was isolated from a soil microbe by Selman A. Waksman, a bacteriologist working in New Jersey. Five years earlier, Waksman had discovered streptomycin, one of medicine's first "miracle drugs," a compound that transformed the treatment of tuberculosis and many other infections. Neomycin--new mycin--was the latest addition to a string of discoveries that ultimately won Waksman a Nobel Prize.

It turned out, however, that neomycin wasn't terribly useful. At high doses it often caused kidney failure and deafness. Today, its use is limited to antibiotic salves and creams. There has been nothing neo about neomycin for ages.

Nothing, that is, until last year, when a biochemist at Harvard Medical School named Guo-fu Hu added neomycin to a culture of human cells. He discovered that it inhibited a protein that is essential to the growth of new blood vessels. That process is crucial to the development of cancer.

Hu and his colleagues are now trying out neomycin in cancer-prone mice in the hope that the nearly forgotten drug--or, more likely, a less toxic version of it--may someday be reborn as an anti-tumor medicine.

"The preliminary experiments are fairly promising," Hu said recently. "I would not be surprised if it inhibits tumor growth."

Meanwhile, a research team in Alabama and another in Philadelphia are experimenting with gentamicin, a chemical cousin of neomycin first used in 1963. Curiously, at the right dose it has the ability to temporarily "correct" certain kinds of genetic mutations. It is being tried in people with cystic fibrosis and Duchenne muscular dystrophy.

That's not all.

French researchers last month reported that diltiazem, a popular "calcium channel blocker" used to lower blood pressure, helps preserve the eyesight of mice with retinitis pigmentosa, an inherited form of blindness. Lovastatin, a wildly successful cholesterol-lowering medicine, is being studied as a cancer preventative. Thalidomide, the notorious sedative that deformed thousands of babies from 1956 to 1963, has been reborn as a treatment for leprosy and multiple myeloma, a bone marrow cancer.

What's going on here is both an old story of medical innovation and a new one.

Unexpected events and insights have always been major engines of scientific discovery. Whether it was Edward Jenner's observation 200 years ago that dairy maids who'd had cowpox never got smallpox, or the more recent observation that an unsuccessful heart drug called sildenafil--Viagra--produced erections in impotent men, serendipity has always been the scientist's very good friend.

The explosion of biological knowledge of the last several decades, however, has both strengthened serendipity's hand and in many ways made it less important in the never-ending effort to find better medicines.

To a degree scarcely imaginable a generation ago, biologists now understand how drugs work. Not all drugs, of course, but many. They understand what receptors they bind to, what cascade of chemical reactions they block or stimulate, how myriad cellular events are changed by them. This knowledge feeds itself. The more scientists learn, the more unopened doors appear before them, and the faster the doors can be opened.

This has led to a new strategy, still in its infancy, of "rational drug design," in which pharmaceutical compounds are constructed from scratch rather than discovered and then altered. At the same time, it has led scientists to look anew at old compounds to better understand their effects (or, more often, their side effects).

"Drugs seldom have a single pharmacological effect. Their multiple effects sometimes cause toxicity, but at other times have therapeutic benefit. There are lots of surprises. But once our knowledge improves, things don't look so surprising," said Alastair J.J. Wood, a clinical pharmacologist at Vanderbilt University.

Sometimes the surprise is that a drug works against a disease in ways nobody expected.

A good example of that is the emerging story of pravastatin, a cholesterol-lowering drug that last year was the 15th biggest prescription drug in sales in the United States. Like others in the "statin" family of compounds, it reduces a person's risk of having a heart attack. Although its effect on cholesterol is the obvious explanation for that, scientists for a while have suspected that there may be others.

For one thing, the benefit of statin drugs occurs almost immediately, long before cholesterol-laden "plaques" in arteries start to shrink. For another, the amount the plaques shrink, even after years of treatment, doesn't seem to be enough to explain the lowered heart attack rate. Perhaps most important, other kinds of drugs don't produce the dramatic effects the statins do--even when they lower cholesterol the same amount.

In recent years, Paul M. Ridker, a cardiologist at Harvard Medical School, and his colleagues have published several studies showing that high levels of C-reactive protein (CRP), a substance produced by inflammation, increases a person's risk of heart attack, just as high cholesterol does.

In July, the research team published a paper showing that in a study from early in the decade, people taking pravastatin had consistently lower CRP levels in their bloodstream than people taking placebo. Furthermore, even in people on placebo who were able to lower their cholesterol (by diet, for instance), CRP didn't go down.

The theory that inflammation promotes the growth of plaques has been around for decades. Ridker's work doesn't prove it. It also doesn't prove that pravastatin exerts some of its power by being, in effect, an anti-inflammatory drug. Nor does it answer the question of whether the other statins do the same thing. However, it has raised all of those ideas as hypotheses that can be tested.

If it turns out the inflammation-as-cardiac-risk-factor theory is correct, that will help explain why nearly half of all heart attacks occur in people with normal cholesterol levels. And that's not all.

"The big question becomes: Will there be novel targets both for treatment and prevention [of heart attack]?" Ridker said. "Through these clinical observations, we might uncover new pathways of disease. Maybe we can use this to better target who should be on these drugs over the long run."

Sometimes, however, the surprise Wood talks about is different. It's the discovery that a drug treats an entirely unexpected disease.

That's what happened with the tetracyclines. They were born in 1948 as antibiotics, but have since found use as anti-rheumatics and are now being studied as anti-cancer agents. As is frequently the case in science, both the wrong theories and the right theories proved illuminating.

The idea that tetracycline might be useful in rheumatoid arthritis goes back to the 1950s and the work of Thomas McPherson Brown, a physician who had a clinic in Arlington. He believed the disease was caused by chronic infection by mycoplasma, a kind of bacteria killed by tetracycline (but few other antibiotics). Evidence for the infection was equivocal, and a small study of tetracycline in the 1970s showed no benefit.

In the 1980s, a group of dental researchers in New York found that tetracycline could protect against tooth loss in diabetic mice. The odd thing was, the protection occurred even in animals raised in a germ-free environment. This suggested that the problem wasn't just chronic gum infection, as they had hypothesized. The researchers eventually discovered that tetracycline inhibits several enzymes that break down connective tissue and cartilage, loosening teeth (among other things).

That suddenly shed new light on tetracycline anecdotes involving rheumatoid arthritis, which is a disease characterized by slow erosion of the cartilage cushioning joints. The NIH eventually sponsored a large study that, in 1995, proved that a drug similar to tetracycline does, indeed, relieve symptoms in some people with the disease.

In truth, tetracyclines aren't much more effective than lots of anti-rheumatic drugs. However, their study has been hugely illuminating. In the last few years, the drugs have become powerful tools for probing a whole new family of enzymes, which operate in the shadowy world between cells. The enzymes, it turns out, help regulate many critical events, including the growth and spread of cancer.

A pharmaceutical company named CollaGenex recently took the next logical step in the tetracycline story. It made a compound that inhibits these enzymes like its tetracycline ancestor but has no antibiotic effect. The activities that had so confounded researchers have now been split from each other, like a river forced into two channels by a wide island. The NIH is now recruiting cancer patients to test the compound.

In most cases where an old drug is found to have a new trick, the evidence accumulates slowly. Occasionally, however, there is a particular moment of insight a scientist remembers.

That's the case with Khandan Keyomarsi's report, published last July, about an unknown effect of the drug lovastatin.

Like the other statins, lovastatin decreases the body's production of cholesterol. In 1991, Keyomarsi had shown that the drug also arrests the growth of cancer cells at a particular stage in their cycle. It appeared to do this by increasing two particular proteins, called p21 and p27. Whether any of this had anything to do with lovastatin's well-known effects on cholesterol was unknown.

One day in 1995, Keyomarsi read a journal article about lactacystin, a compound that blocks a cell structure called a "proteasome" that, among other things, chews up compounds like p21 and p27. She noticed that lactacystin's chemical structure bore some resemblance to lovastatin's. Could lovastatin also be attacking proteasomes, causing p21 and p27 to rise?

Keyomarsi doubted it because she knew that the body changed lovastatin into a compound that no longer looked like lactacystin. But what if some of the original escaped that change? It was a question worth answering.

She went across the hall from her office at the Wadsworth Center, a state-supported research laboratory in Albany, N.Y., and took out of the freezer a sample of lovastatin that had been metabolized by the body. She put it in a machine that determined how many distinct chemicals it contained. The answer: two. In fact, about one-fifth of the lovastatin was unchanged.

"In this case, we got lucky. We were looking for a connection, and we found it," Keyomarsi said. "We saw the paper in the morning, and by the afternoon we knew we were on the right track."

She has gone on to show that lovastatin, indeed, affects cancer cells in the way lactacystin does--which is entirely distinct from how it affects the body's production of cholesterol. Whether this will lead to the use of lovastatin--or something like it--in cancer prevention is, of course, unknown. But now it is possible.